You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hive.apache.org by "Namit Jain (JIRA)" <ji...@apache.org> on 2011/02/03 01:06:28 UTC

[jira] Created: (HIVE-1949) Code cleanup: ExecMapperContext.inputFileChanged() should only be called by ExecMapper/MapOperator

Code cleanup: ExecMapperContext.inputFileChanged() should only be called by ExecMapper/MapOperator
--------------------------------------------------------------------------------------------------

                 Key: HIVE-1949
                 URL: https://issues.apache.org/jira/browse/HIVE-1949
             Project: Hive
          Issue Type: Improvement
            Reporter: Namit Jain


Currently, inputFileChanged() will only return true for the first row.
Consider the following scenario:

select /*+ mapjoin(B) */ ... from A join B where A.c1 < 100;

If the first row from A > 100, then the MapJoinOperator will never receive the first row, 
and if MapJoinOperator calls inputFileChanged() - that function will be called for the 
second row, and therefore return false.

It would be cleaner for ExecMapper/MapOperator to check whether input file changed or not,
and then call some new function for each operator informing them that the input file has 
changed.

-- 
This message is automatically generated by JIRA.
-
For more information on JIRA, see: http://www.atlassian.com/software/jira